1. Field of the Invention
The present invention relates to a projection system for a charged particle multi-beamlet system, such as for a charged particle multi beamlet lithography system or an inspection system, and an end module for such a projection system.
2. Description of the Related Art
Currently, most commercial lithography systems use a mask as a means to store and reproduce the pattern data for exposing a target, such as a wafer with a coating of resist. In a maskless lithography system, beamlets of charged particles are used to write the pattern data onto the target. The beamlets are individually controlled, for example by individually switching them on and off, to generate the required pattern. For high resolution lithography systems designed to operate at a commercially acceptable throughput, the size, complexity, and cost of such systems becomes an obstacle.
One type of design used for charged particle multi-beamlet systems is shown for example in U.S. Pat. No. 5,905,267, in which an electron beam is expanded, collimated and split by an aperture array into a plurality of beamlets. The obtained image is then reduced by a reduction electron optical system and projected onto a wafer. The reduction electron optical system focuses and demagnifies all the beamlets together, so that the entire set of beamlets is imaged and reduced in size. In this design, all the beamlets cross at a common cross-over, which introduces distortions and reduction of the resolution due to interactions between the charged particles in the beamlets.
Designs without such a common cross-over have also been proposed, in which the beamlets are focused and demagnified individually. However, when such a system is constructed having a large number of beamlets, providing multiple lenses for controlling each beamlet individually becomes impractical. The construction of a large number of individually controlled lenses adds complexity to the system, and the pitch between the lenses must be sufficient to permit room for the necessary components for each lens and to permit access for individual control signals to each lens. The greater height of the optical column of such a system results in several drawbacks, such as the increased volume of vacuum to be maintained and the long path for the beamlets which increases e.g. the effect of alignment errors caused by drift of the beamlets.
A charged-particle multi-beam exposure apparatus is disclosed in patent publication US2001/0004185, for exposure of a target. An illumination system is adapted to produce an electron beam and form it into a substantially telecentric beam illuminating an element electron optical system array comprising an aperture array, blanker array, and stopper array. The electrons propagate along separate paths through the element electron optical system. A projection optics system projects the resulting beamlets onto the target. The illuminating system and the projection optics system use particle-optical lenses with lens elements common to more than one electron beamlet, so-called macro optical elements.
The apparatus according to this state of the art typically comprises a sequence of two so-called symmetrical magnetic doublets, which produce two cross-overs common to the beamlets within a set of telecentric beams or beamlets. While in principle multiple columns of the type described above may be used side by side to increase capacity, the size of the lens systems involved in this type of system make this solution impractical.
The use of macro optical elements as known from these early designs for charged particle systems, does not permit the beamlets to travel straight throughout the charged particle column and makes uniform control of the beamlets more difficult. For systems using a very large number of particle beamlets, such designs are thus considered sub-optimal in view of ultra-high precision requirements for operations such as overlay and stitching. Applying elements for individual adjustment of a large number of beamlets is moreover considered overly complicated.
To meet the demand for ever smaller nodes (the next node being defined as a factor 1.4 or sqrt(2) smaller dimensions), it is required to reduce the spot size by a factor of 1.4 in charged particle systems, and to double the total current in the system. When reducing spot size by a factor of 1.4, a smaller point spread function (PSF) of charged particle beamlets is required to maintain exposure latitude. The current per beamlet drops by a factor of four due to the smaller PSF, given that the current per beam is defined by a system constant C, a source brightness, and the fourth power of the PSF. The total current on the target, in many cases a wafer, should double to limit shot noise, so that the number of electrons per square critical diameter remains the same, while the resist sensitivity should double, e.g. from 30 to 60 □m per cm2. All in all, so as to maintain throughput, the amount of additional current required to realize a next technical node in charged particle lithography, requires an eight-fold increase in the combined effect of system constant C, the number of beams in the system, and source brightness. While system constant and source brightness may be varied only to a limited extent, the number of beamlets in a system can be considerably increased. In a practical implementation processing wafers as a target, and achieving a throughput of at least e.g. ten wafers per hour, the number of beamlets required is in the order of tens of thousands to hundreds of thousands.
One system addressing such need for a vast multiplicity of beamlets (beamlets per square surface), and also addressing the difficulties encountered in dealing with aberrations as encountered in common cross-overs as in the prior art, is known from patent U.S. Pat. No. 6,958,804 in the name of present applicant. The lithography system defined by this patent allows the inclusion of a vast multiplicity of beamlets maintained on separate paths, i.e. without a common cross-over, by applying arrayed charged particle optical members, such as electrostatic elements, virtually throughout the charged particle system, including the projection parts. This principle difference in layout of the charged particle column of the system allows proper control of all of the beamlets in the system and does not require specific adaptations in response to differences in field strength over the cross section of macro components such as a macro deflector as required in the prior art. Moreover, the application of arrayed elements at least more easily allows application of high frequency switching, which is difficult if not impossible at macro components such as macro-deflectors. Yet even this technology is faced with limits as to number of projection lens systems per square surface, in that at ultimate miniaturization and close distribution of lens systems per square surface, practical problems are encountered in flash over of electric fields of subsequent projection lens elements in the charged particle column.
Where existing charged particle beam technology is suitable for lithography systems for relatively course patterning of images, for example to achieve critical dimensions of 90 nm and higher, a growing need exists for improved performance. It is desired to achieve considerably smaller critical dimensions, for example 22 nm, while maintaining sufficient wafer throughput, e.g. between 10 and 60 wafers per hour.